Journal of Animal Blackwell Publishing Ltd. Ecology 2003 Web-location by linyphiid spiders: prey-specific 72, 745–756 aggregation and foraging strategies

JAMES D. HARWOOD*, KEITH D. SUNDERLAND† and WILLIAM O. C. SYMONDSON* *School of Biosciences, Cardiff University, PO Box 915, Cardiff CF10 3TL, UK; and †Horticulture Research International, Wellesbourne, Warwick CV35 9EF, UK

Summary 1. Web-location cannot be designed simply to optimize foraging success, but must be a compromise between competing factors, including microhabitat parameters (physical structure, microclimate) and predator avoidance. 2. We tested the hypothesis that, despite these compromises, linyphiid spiders managed to locate their webs within microsites with enhanced prey resources. 3. We tested further the ‘equilibrium’ hypothesis that spiders, within a community of species, occupy different, relatively narrow niche axes and despite being generalists do not simply aggregate to the same resources. 4. In the first study of its kind, prey resources at web-sites were compared with those at matched non-web-sites with the same microhabitat properties. This was carried out for two subfamilies of Linyphiidae, one of which (Linyphiinae) is web-dependent and locates its web just above the ground, and the other (Erigoninae) is less web-dependent, locating its web on the ground and often hunting away from its web. We hypothesized that these subfamilies would locate their webs at microsites with enhanced densities of the prey most likely to be captured by their contrasting foraging strategies. We used spider weights as a surrogate for nutritional state, and compared spiders in webs with those without webs. 5. We found that even within a relatively uniform crop of wheat, spiders located their webs at microsites with greater prey resources. The most numerous prey, Collembola, were at significantly greater density at web-sites of both subfamilies of spider. However, there were significantly more Collembola at web-sites of the surface-hunting Erigo- ninae. By contrast, significantly more non-Collembola prey, especially aphids and Thysanoptera, were found at web-sites of the Linyphiinae, whose aerial webs might be expected to intercept such prey more efficiently. 6. These differences were also present at matched non-web-sites, suggesting that micro- site selection by the spiders was determined by the spectra of prey present. Comparison of the weights of web-owning and non-web-owning Tenuiphantes tenuis (Linyphiinae) demonstrated the nutritional benefits of web-owning. 7. The results suggested that the spiders used a combination of web-location strategies at the microhabitat level, foraging behaviour and (known) high-intensity intraspecific competition, to exploit prey-rich microsites efficiently within fields in a dynamic manner. Key-words: alternative prey, Linyphiidae, microsite selection, niche axes, optimal foraging. Journal of Animal Ecology (2003) 72, 745–756

fitness (e.g. Charnov 1976; Pyke 1984; Stephens & Krebs Introduction 1986). However, as highlighted by Pierce & Ollason Foraging strategies by animals have been studied exten- (1987), foraging is rarely a wholly independent activity; sively in relation to maximizing energy intake and independent, that is, from the need to evade predators, to find a mate or to hold onto a safe refuge, for example. Correspondence: Dr J. D. Harwood, School of Biosciences, This lack of independence, but strong interdependence, © 2003 British Cardiff University, PO Box 915, Cardiff CF10 3TL, UK. has been used to attack practical applications of opti- Ecological Society Fax: + 44 (0) 29 20874305; E-mail: [email protected] mal foraging models. We wished to test the hypothesis

746 that, despite these potential sources of error, a group local aggregation to prey density would be determined J. D. Harwood, of generalist predators, linyphiid spiders, living within by the types of prey most readily captured by the con- K. D. Sunderland & a relatively homogeneous ecosystem (a crop of wheat), trasting foraging strategies of the spiders. Thus it might W. O. C. Symondson were managing to locate their webs at sites where prey be predicted that web-dependent species (Linyphiinae) were aggregated at the microhabitat level. We further would locate their webs where they were likely to cap- wished to test the hypothesis that this web-location ture more active flying insects, while species that also behaviour, although limited by powerful constraints on hunt over the soil surface (Erigoninae) might be affected the type of sites suitable for the construction of webs, strongly by the densities of less mobile epigeal prey. resulted in an improved nutritional status for the spiders. Harwood, Sunderland & Symondson (2001) showed We wished to establish whether or not this would be detec- that linyphiid spiders were not locating their webs at table despite the need to satisfy other constraints on random within wheat fields but selecting web-sites where web-location imposed by microhabitat properties (tem- prey density was higher than at random non-web-sites. perature, humidity, light levels, protection from wind, However, even within superficially uniform wheat fields etc.) and the need to satisfy other survival parameters, there is considerable microhabitat heterogeneity. For including predator avoidance and mate finding. example, webs might be located within rows of wheat Marginal value theories (Charnov 1976) can be plants or between them. Such spatial heterogeneity affects applied to patch residence time only when the quality of a both the suitability of microsites (defined as a potential site is reduced continuously and predictably by patch web-site within a microhabitat) for web-location depression through predation (Stephens & Krebs 1986). (which in turn will be different for different spider sub- However, with linyphiid spiders there is continuous families and species) and the distribution of prey that inter- and intraspecific competition for web-sites, with are similarly responsive to microhabitat variables (Adams rapid displacement of one spider by another. For exam- 2000). In this study, therefore, we based our prey sam- ple, adult female Linyphiinae, such as Tenuiphantes pling protocol upon pairwise matching of spider web-sites tenuis (Blackwall) (= Lepthyphantes tenuis), compete with non-web-sites possessing the same microhabitat strongly for web-sites, with larger spiders evicting smaller properties. This allowed us to test the hypothesis that ones, resulting in a mean residence time for an indi- the spiders and prey were simply responding to the vidual spider in a web-site of below 2 days (Samu et al. same microhabitat cues; should this be occurring then 1996). Thus the residence time of an individual spider web and matched non-web-sites should have the same in a web-site (as distinct from choice of web-site) may mean density of prey, or there could even be fewer prey be less to do with site quality and more with the domin- at web-sites through prey depletion. Should the Liny- ance hierarchy within a spider population or commu- phiinae and Erigoninae be responding spatially to dif- nity. Although it might be assumed that spiders will ferent types of prey then an analysis of prey density at change their web-sites when prey availability becomes web-sites of these two subfamilies should reveal such inadequate, Smallwood (1993) showed that, paradoxic- differences, which might suggest resource partitioning ally, long-jawed spiders were more likely to relocate and hence avoidance of competition. We also wished to their webs where prey density was high as a result of test the hypothesis that the different microsites to which increased intraspecific competition between the spi- the different spider subfamilies were responding would ders. The residence time of a web-site, rather than of a show differences in the relative abundance of different particular spider at a particular web-site, may therefore types of prey. Comparison of prey density and type at be a more relevant indicator of site quality. Whether or non-web-sites, which were matched structurally to the not a site is abandoned will depend on the density of web-sites of the Linyphiinae and Erigoninae but not competing sites of equal or better quality. However, chosen by them as web-sites, provided a means of testing whether depletion of prey at web-sites occurs will also this hypothesis. depend upon both the hunting strategy of the predator The combination of exploitation of structurally dif- and the mobility of the prey. For example, among the ferent microhabitats within the same arable ecosystem Linyphiidae there are two subfamilies. Erigoninae con- (including a degree of vertical stratification) with dif- struct small sheet webs in the horizontal plane on or ferences in hunting strategies may allow coexistence very near to the ground (Sunderland, Fraser & Dixon between different species and subfamilies of Linyphiidae. 1986; Alderweireldt 1994a), which are used both for direct Such separation of species may have evolved as a mech- capture of prey and as a base from which to actively anism permitting close association within adjacent pursue prey (Alderweireldt 1994a). In contrast, Liny- but more spatially confined niche axes, reducing inter- phiinae build relatively large horizontal webs a few specific competition. Spider communities can be remark- centimetres above the ground (Sunderland et al. 1986; ably diverse in wheat fields (Sunderland, Chambers & Alderweireldt 1994a), using vegetation for the attach- Carter 1988). There are many parallels between the ment of silk. Linyphiinae rarely hunt away from the organization of plant communities and communities of © 2003 British web and are strongly reliant upon it for the capture of web-building spiders in habitats such as species-rich hay Ecological Society, Journal of Animal prey (Alderweireldt 1994a). Thus, two closely related meadows. Despite Gause’s competitive exclusion principle Ecology, 72, groups of spiders have evolved simple but contrasting (Crawley 1997), ecologically similar plant species do 745–756 foraging strategies. We tested the hypothesis that any coexist in such habitats, as do a range of web-building

747 spiders which share similar areas of resource space. falling and flying invertebrates. A ground search within Spider There are many competing theories for coexistence of mini-quadrats (78·5 cm2) provided an instantaneous web-location plant species, both equilibrium and stochastic (reviewed measure of arthropod abundance which, although less strategies in Crawley 1997; Silverton & Charlesworth 2001). efficient at monitoring flying insects, was more accurate Different spider species may each favour a particular for recording densities of less active invertebrates. microhabitat for web construction (just as different Weather conditions may also affect the efficiency with plant species may favour particular microhabitats), or which each trapping system operates, as the activity and they may respond to a combination of different con- movement patterns of insects can change in response to ditions (equilibrium theories). Alternatively, a less temperature, humidity or light levels. We assumed that competitive species may survive in the presence of a if either trapping system showed a significant differ- more competitive species in a fluctuating environment ence between treatments, then that represented a reli- if, for example, the less competitive species does better able record of invertebrate abundance, even if the other at some point in the fluctuation cycle or is better at dis- trapping system showed no significant differences. Mini- persal (non-equilibrium theories). Many other factors, sticky traps and mini-quadrats are henceforth referred including predation (of the spiders) or existence of ref- to as ‘sticky traps’ and ‘quadrats’. ugia (from more competitive species) may be involved. Spiders were collected at random, either from webs or Here, however, we test the theory that the spiders occupy from the parts of the ground surface or on the vegetation different, relatively narrow niche axes and, despite being where webs were not present. They were collected by generalists, do not simply compete for, and aggregate pooter and placed into separate 1·5 mL Eppendorf to, the same resources. tubes on ice. The spiders were killed by freezing at It is reported commonly that a high proportion of −20 °C within 1 h of collection and categorized as either individuals in spider populations are frequently in a ‘web-owners’ if they were collected from within a web, or state of semi-starvation (Anderson 1974; Riechert 1992; ‘non-web-owners’ if found elsewhere. Non-web-owners Bilde & Toft 1998) and that food availability can often were collected from areas of wheat that were not be a limiting factor to population density (Brown 1981; included in the matched microhabitat sampling (see Wise 1993). If the food supply is also spatially and tem- below). If more than one spider occupied a single web porally heterogeneous there will be strong selection all spiders present were categorized as ‘web-owners’ and pressures on web-building spiders to evolve efficient were assumed to have occupied a single web-site. Webs mechanisms for locating and defending prey-rich containing spiders were paired with non-web-sites which web-sites, and develop a behavioural repertoire of dynamic were chosen by selecting a location as visually similar relocation in relation to shifting resource levels (Janetos as possible to the web-site in terms of microhabitat 1982; Olive 1982; Gillespie & Caraco 1987; Uetz 1992). structure. For example, if a web was within a row of wheat Where prey resources are good the optimal strategy stems, a paired non-web-site (up to 30 cm away) was would appear to be not to relocate if it can be avoided selected that was also within a row of structurally (Leclerc 1991). Spiders that have lost their webs, similar wheat stems, but which contained no web. or been ousted by more dominant individuals, would Sampling, using sticky traps or quadrats, was always be expected to be at a nutritional disadvantage. We carried out in pairs to enable direct comparisons wished to test the hypothesis that this would affect between prey densities at web and non-web-sites. the web-dependent Linyphiinae more strongly than the Unless stated otherwise, sampling was as described in less web-dependent Erigoninae. This could be done by Harwood et al. (2001). The four fields were always weighing all spiders in the field study and interpreting sampled on the same day using both monitoring tech- weights in terms of recent feeding history, by reference niques (n = 15 paired samples per sampling technique to spiders of known nutritional status in the laboratory per date per field, on seven sampling dates). (Riechert 1992).      Materials and methods 

  Ten individuals of each potential prey species were col- lected from the field and weighed. The mean weight for The study was carried out in four fields (7–14 ha) of each species was used to calculate total prey biomass winter wheat at Horticulture Research International, collected during sampling. Spiders collected by pooter Warwickshire, UK. were also weighed because spider mass is known to be a Linyphiid spiders were collected every 2 weeks from good indicator of nutritional condition (Nakamura 1977). early May until harvest (late July 1999). Two comple- mentary sampling techniques were used and described © 2003 British      in detail by Harwood et al. (2001). Mini-sticky traps Ecological Society,  2 T. TENUIS Journal of Animal (7·5 cm ) were placed on the ground and left in situ for Ecology, 72, 24 h, providing a cumulative measure of invertebrate Male and female T. tenuis were collected from the field 745–756 activity-density; these traps collect ground-active, and placed individually in 50 × 15 mm triple-vented

748 Petri dishes containing a plaster of Paris and charcoal J. D. Harwood, base to ensure high humidity. Spiders were provided K. D. Sunderland & ad libitum with live Drosophila melanogaster Meigen W. O. C. Symondson (Diptera: Drosophilidae) and Isotoma anglicana (Lubbock) (Collembola: Isotomidae) for 24 h, transferred into new containers and starved for 2 weeks. Fifteen spiders of each sex were allocated randomly to six feed- ing regimes (16 °C, 16L : 8D). Spiders were fed one D. melanogaster every day, or every 2nd, 3rd, 4th, 6th or 12th day. All spiders were weighed before the start of the feeding trial, on day 12 (prior to feeding), day 13 (one day after feeding) and again on days 24, 25, 36 and 37. If at any stage the spider died, data collected from those individuals at earlier time periods were excluded.

 

The classification of arthropods as potential prey for linyphiid spiders was based on data from published literature and prey acceptability trials (reviewed in Harwood et al. 2001). Throughout the text any reference to ‘prey’ refers to acceptable prey items only. All arthropod population data (numerical and bio-

mass) were log10(x + 1)- or sqrt(x + 1)-transformed prior to analysis by paired sample t-tests or . The number and biomass of individual prey taxa (Collembola, Diptera, Hymenoptera, Coleoptera, Aphididae and Thysanoptera) expressed as a proportion of the total prey captured were arcsine-transformed. When variances of less common prey taxa could not be stabilized, non-parametric analyses were performed.

Results Spiders captured in winter wheat fields were dominated by the linyphiid subfamilies Erigoninae (n = 374 web- owners and n = 256 non-web-owners) and Linyphiinae (n = 724 web-owners and n = 291 non-web-owners). The Erigoninae consisted primarily of three species, Erigone atra (Blackwall) (n = 321), E. dentipalpis (Wider) Fig. 1. Differences between prey at web (grey bars) and non- (n = 85) and Oedothorax spp. (n = 112), and the Liny- web (white bars) sites of linyphiid spiders (*P < 0·05, phiinae were dominated by T. tenuis (n = 612) and **P < 0·01 and ***P < 0·001). Means (± SE) presented per 2 Bathyphantes gracilis (Blackwall) (n = 205). cm for number of prey captured on sticky traps (a) and quadrats (b), biomass [µg] of prey captured on sticky traps (c) and quadrats (d). Mean number of Araneae are also presented   and corrected to x + 1 to account for the web owner in web-centred quadrat catches. Biomass of Araneae are not Numbers of prey captured are shown in Fig. 1. The presented. relative importance of different prey taxa depended upon whether numbers or biomass were analysed. For whether monitored by sticky traps (mean per web-site ± ± example, as proportions of total prey captured within 3·54 0·21, per non-web-site 2·45 0·14) (F1,784 = 48·76, quadrats, Collembola constituted 0·85 of prey items P < 0·001) or quadrats (mean per web-site 13·12 ± 0·68, ± numerically, but only 0·58 of prey biomass. Both Dip- per non-web-site 7·48 0·49) (F1,784 = 109·94, P < 0·001). tera and Aphididae were consistently a more important Similarly, web-sites contained a significantly greater food resource, in terms of biomass, than their numbers biomass of prey compared to paired non-web-sites on suggested, while the smallest prey, Thysanoptera and both sticky traps (mean per web-site 0·99 ± 0·06 mg, © 2003 British Collembola, were less so. per non-web-site 0·69 ± 0·04 mg) (F = 31·60, P < Ecological Society, 1,784 ± Journal of Animal Importantly, when linyphiid web and non-web-sites 0·001) and within quadrats (mean per web-site 2·82 ± Ecology, 72, were matched for microhabitat structure, the number 0·14 mg, per non-web-site 1·58 0·10 mg) (F1,784 = 102·34, 745–756 of prey items within web-sites was significantly greater, P < 0·001).

749 Significantly more Collembola, Coleoptera, Aphidi- web-sites of Erigoninae than at those of Linyphiinae. Spider dae and Araneae were captured on sticky traps within The only other group that were more numerous at the web-location web-sites compared to matched non-web-sites nearby web-sites of Erigoninae, than at non-web-sites, were strategies (Fig. 1). However, of these, only the biomass of Colle- Coleoptera (Table 1). By contrast, Coleoptera, Aphidi- mbola was significantly greater at web-sites compared dae, Hymenoptera, Thysanoptera and Diptera were all to non-web-sites. The only instance where there were significantly more abundant in Linyphiinae web-sites significantly fewer individuals captured at web-sites was than in non-web-sites using one or both trapping sys- for Hymenoptera, but their mean biomass was greater tems (Table 2). Overall, quadrat data showed that

at web-sites. Quadrat data indicated that the number there were significantly greater numbers (F1,438 = 5·52,

and biomass of Diptera (as well as Collembola, Cole- P < 0·05) and biomass (F1,438 = 8·94, P < 0·01) of optera and Aphididae) were significantly greater at non-Collembola prey at the web-sites of Linyphiinae web-sites as well. compared with Erigoninae. Sticky traps showed no Despite the fact that few Collembola would be avail- significant differences. able at the level of Linyphiinae webs, there were never- Despite these differences, the number of prey were theless more Collembola (and total prey) at both similar at web-sites of the two spider subfamilies. How- Linyphiinae and Erigoninae web-sites than at matched ever, significantly more Collembola were captured at non-web-sites when these subfamilies were analysed web-sites of Erigoninae compared to those of Linyph- separately (Tables 1 and 2). The comparisons were sig- iinae (Table 3). Among the less common prey, signi- nificant whether numbers or biomass were analysed ficantly more Aphididae and Thysanoptera were present and regardless of trapping system. Interestingly, how- at web-sites of Linyphiinae when monitored by sticky ever, there were up to twice as many Collembola at traps (Aphididae: Mann–Whitney U = 15919·5, P < 0·001;

Table 1. Difference between web and non-web-sites of Erigoninae spiders in the number and biomass of prey captured by sticky traps and within quadrats during 1999. Mean number of Araneae captured in web-centred quadrats are corrected to x + 1 to account for the web-owner

Mean per Mean per non- Ratio Variable tP web-site ± SE web-site ± SE (web/non-web)

(a) Sticky traps: prey number (n = 109) Collembola 3·10 0·002 3·07 ± 0·48 1·99 ± 0·30 1·54 Diptera 0·30 0·762 0·22 ± 0·05 0·20 ± 0·04 1·09 Hymenoptera 0·91 0·363 0·08 ± 0·03 0·13 ± 0·04 0·64 Coleoptera 3·38 0·001 0·20 ± 0·05 0·02 ± 0·01 11·03 Aphididae 1·63 0·104 0·49 ± 0·16 0·28 ± 0·08 1·71 Thysanoptera 0·73 0·465 0·17 ± 0·04 0·21 ± 0·05 0·78 Araneae 1·89 0·059 0·12 ± 0·03 0·05 ± 0·02 2·60 Total prey 3·77 < 0·001 4·30 ± 0·52 2·86 ± 0·33 1·50 (b) Sticky traps: prey biomass (mg) (n = 109) Collembola 3·56 < 0·001 0·61 ± 0·10 0·38 ± 0·06 1·59 Diptera 0·04 0·967 0·12 ± 0·03 0·12 ± 0·03 1·01 Hymenoptera 0·75 0·455 0·05 ± 0·01 0·04 ± 0·01 1·37 Coleoptera 1·80 0·072 0·09 ± 0·03 0·02 ± 0·01 4·60 Aphididae 1·08 0·280 0·28 ± 0·09 0·17 ± 0·05 1·62 Thysanoptera 0·73 0·465 0·01 ± < 0·01 0·01 ± < 0·01 0·79 Total prey 3·71 < 0·001 1·21 ± 0·15 0·76 ± 0·09 1·58 (c) Quadrats: prey number (n = 124) Collembola 7·56 < 0·001 15·23 ± 1·74 8·72 ± 1·15 1·75 Diptera 1·34 0·180 0·10 ± 0·03 0·06 ± 0·02 1·85 Hymenoptera 0·94 0·346 0·10 ± 0·03 0·06 ± 0·02 1·50 Coleoptera 2·91 0·004 0·25 ± 0·05 0·09 ± 0·03 2·81 Aphididae 0·74 0·460 0·41 ± 0·14 0·33 ± 0·13 1·24 Thysanoptera 1·34 0·179 0·21 ± 0·05 0·13 ± 0·03 1·63 Araneae 9·93 < 0·001 1·34 ± 0·07 0·09 ± 0·03 15·54 Total prey 7·61 < 0·001 17·79 ± 1·83 10·22 ± 1·20 1·74 (d) Quadrats: prey biomass (mg) (n = 124) Collembola 5·45 < 0·001 2·66 ± 0·31 1·60 ± 0·22 1·66 Diptera 1·04 0·300 0·05 ± 0·01 0·03 ± 0·01 1·88 Hymenoptera 1·50 0·133 0·03 ± 0·01 0·01 ± 0·01 1·74 © 2003 British Coleoptera 2·55 0·011 0·24 ± 0·05 0·09 ± 0·03 2·68 Ecological Society, Aphididae 1·02 0·308 0·27 ± 0·09 0·20 ± 0·07 1·32 Journal of Animal Thysanoptera 1·34 0·179 0·01 ± < 0·01 0·01 ± < 0·01 1·57 Ecology, 72, Total prey 6·21 < 0·001 3·44 ± 0·34 2·06 ± 0·23 1·67 745–756 750 Table 2. Difference between web and non-web-sites of Linyphiinae spiders in the number and biomass of prey captured by sticky J. D. Harwood, traps and within quadrats during 1999. Mean number of Araneae captured in web-centred quadrats are corrected to x + 1 to K. D. Sunderland & account for the web-owner W. O. C. Symondson Mean per Mean per non- Ratio Variable tP web-site ± SE web-site ± SE (web/non-web)

(a) Sticky traps: prey number (n = 317) Collembola 5·57 < 0·001 1·92 ± 0·22 1·27 ± 0·14 1·52 Diptera 1·38 0·168 0·23 ± 0·03 0·18 ± 0·03 1·29 Hymenoptera 5·92 < 0·001 0·05 ± 0·01 0·26 ± 0·03 0·20 Coleoptera 6·50 < 0·001 0·29 ± 0·04 0·04 ± 0·01 7·15 Aphididae 3·11 0·002 0·46 ± 0·08 0·27 ± 0·05 1·69 Thysanoptera 0·92 0·360 0·21 ± 0·03 0·24 ± 0·03 0·87 Araneae 4·98 < 0·001 0·16 ± 0·03 0·03 ± 0·01 5·77 Total prey 5·47 < 0·001 3·25 ± 0·23 2·29 ± 0·15 1·42 (b) Sticky traps: prey biomass (mg) (n = 317) Collembola 3·82 < 0·001 0·35 ± 0·04 0·24 ± 0·03 1·46 Diptera 0·63 0·526 0·13 ± 0·02 0·11 ± 0·02 1·18 Hymenoptera 1·21 0·226 0·08 ± 0·01 0·06 ± 0·01 1·27 Coleoptera 0·28 0·776 0·05 ± 0·01 0·05 ± 0·01 1·05 Aphididae 1·09 0·274 0·22 ± 0·04 0·16 ± 0·03 1·38 Thysanoptera 0·92 0·360 0·01 ± < 0·01 0·01 ± < 0·01 0·88 Total prey 4·18 < 0·001 0·91 ± 0·07 0·66 ± 0·05 1·38 (c) Quadrats: prey number (n = 316) Collembola 11·36 < 0·001 9·54 ± 0·69 5·66 ± 0·52 1·69 Diptera 3·94 < 0·001 0·13 ± 0·02 0·04 ± 0·01 3·10 Hymenoptera 1·47 0·142 0·10 ± 0·02 0·06 ± 0·02 1·49 Coleoptera 6·58 < 0·001 0·38 ± 0·04 0·09 ± 0·02 4·24 Aphididae 3·01 0·003 0·81 ± 0·10 0·54 ± 0·08 1·48 Thysanoptera 2·14 0·033 0·52 ± 0·08 0·33 ± 0·04 1·58 Araneae 15·66 < 0·001 1·46 ± 0·04 0·07 ± 0·01 21·80 Total prey 13·19 < 0·001 13·00 ± 0·75 7·86 ± 0·54 1·77 (d) Quadrats: prey biomass (mg) (n = 316) Collembola 4·70 < 0·001 1·51 ± 0·12 0·95 ± 0·09 1·60 Diptera 4·11 < 0·001 0·07 ± 0·01 0·02 ± 0·01 3·54 Hymenoptera 2·25 0·024 0·03 ± 0·01 0·02 ± < 0·01 1·75 Coleoptera 6·36 < 0·001 0·36 ± 0·04 0·08 ± 0·02 4·19 Aphididae 3·21 0·001 0·60 ± 0·08 0·37 ± 0·05 1·63 Thysanoptera 2·14 0·033 0·02 ± < 0·01 0·02 ± 0·01 1·53 Total prey 9·65 < 0·001 2·83 ± 0·16 1·59 ± 0·11 1·78

Table 3. Statistical comparison of prey captured at web and matched non-web-sites of erigonine spiders vs. web and matched non-web-sites of linyphiine spiders. Sticky trap and quadrat data were analysed separately. Values compared are presented in Tables 2 and 3

Variable Sticky trap Quadrat

(a) Web-sites

Prey number F1,424 = 2·05, P = 0·153 F1,438 = 2·66, P = 0·103

Collembola number F1,424 = 5·44, P = 0·020 F1,438 = 10·13, P = 0·002

Prey biomass F1,424 = 4·62, P = 0·032 F1,438 = 3·42, P = 0·065

Collembola biomass F1,424 = 7·44, P = 0·007 F1,438 = 17·82, P < 0·001 (b) Non-web-sites

Prey number F1,424 = 0·84, P = 0·359 F1,438 = 3·32, P = 0·069

Collembola number F1,424 = 7·55, P = 0·006 F1,438 = 7·87, P = 0·005

Prey biomass F1,424 = 1·19, P = 0·227 F1,438 = 4·28, P = 0·039

Collembola biomass F1,424 = 5·93, P = 0·015 F1,438 = 10·68, P = 0·001

Thysanoptera: U = 17525·5, P < 0·001) and quadrats subfamilies of spider on sticky traps (Aphididae: © 2003 British (Aphididae: U = 15880·0, P < 0·001; Thysanoptera: U = 16835·0, P < 0·01; Thysanoptera: U = 17148·0, Ecological Society, Journal of Animal U = 17525·0, P < 0·05). There were also significant dif- P < 0·01) and quadrats (Aphididae: U = 17148·0, Ecology, 72, ferences between the numbers of these two prey taxa P < 0·01; Thysanoptera: U = 16822·0, P < 0·01). The 745–756 captured in the matched non-web-sites of these two only other less common prey to differ between sites of 751 the two subfamilies were Hymenoptera whose number Spider were significantly higher at non-web-sites of the Erig- web-location oninae compared to non-web-sites of Linyphiinae strategies (U = 15482·5, P < 0·05). This could be because, in wheat, hymenopterous eggsac parasitoids of Erigoni- nae are more prevalent than those attacking eggsacs of Linyphiinae (Van Baarlen, Sunderland & Topping 1994). To ensure that the processes we were observing were widespread, data were collected by sampling in four fields with different histories. Not surprisingly, there were large between-field differences in the number and biomass of prey at web-sites ( P < 0·001). The interactions between location (web or non-web-site) and field were not significant, indicating that the same processes were occurring in all fields. The mean number of prey items captured within the field containing the greatest density of prey was 3·6 × and 3·9 × greater on sticky traps and within quadrats, respectively, compared to the field with the lowest prey density.

  

The weight of spiders in different regimes varied signi- ficantly after days 12, 24 and 36 days for both females and males (Fig. 2) ( P < 0·001). The mean weight of laboratory-reared D. melanogaster [0·95 ± 0·03 mg (n = 20)] enabled the mean biomass consumed per spi- der per day within each of the six feeding regimes to be estimated. Regression equations for spider weight (x)

after 36 days against log10 prey biomass available (y) for female and male spiders were as follows: − 2 Female: y = 0·35x 1·32; r = 0·68, F1,79 = 171·29, P < 0·001 ± Male: y = 1·29x − 2·31; r2 = 0·86, F = 445·50, P < Fig. 2. Mean weight ( SE) of (a) female and (b) male Tenui- 1,74 phantes tenuis subjected to six different feeding regimes. Spiders 0·001 were weighed before the experiment (day 0), and before feeding on days 12, 24 and 36. Spiders in regimes 1–6 were fed    - one fly every 1, 2, 3, 4, 6 and 12 days, respectively. 

Spiders were weighed to determine the relationship was dependent upon, and related positively to, avail- between prey availability in the field and spider weight. ability of prey at web-sites. This was despite high levels E. atra, E. dentipalpis, T. tenuis and B. gracilis were cap- of background noise in the regression due to sampling tured in sufficient numbers from web and random non- across four fields of varying prey density on seven web-sites to enable body weight to be analysed (Fig. 3). different dates. Only web-owning T. tenuis were found to be signi- Using the equations derived above, it was possible to ficantly heavier than non-web-owners of the same estimate total prey consumption by field-collected male

species (female: F1,312 = 33·71, P < 0·001; male: F1,147 = and female T. tenuis from their body weights, since 9·83, P < 0·01). The weight of spiders collected during average summer temperatures in UK wheat fields are a separate study in 1998 (Harwood et al. 2001) showed comparable to those used for the laboratory experi- a similar species-specific effect for T. tenuis (female: ments (Harwood et al. 2001). Given that estimated

F1,212 = 13·97, P < 0·001; male: F1,46 = 5·28, P < 0·05). daily consumption by female T. tenuis (mean estimated We analysed first the relationship between spider consumption = 0·65 ± 0·04 mg day−1) was significantly ± −1 weight and two dependent variables, prey number and greater than by males (0·19 0·01 mg day ) (F1,461 = ‘field’, with the latter set as a categorical variable. The 120·46, P < 0·001), analyses were conducted separately ‘field’ variable did not improve r2 values and was for the two sexes. The mean estimated consumption © 2003 British removed. A highly significant correlation was found by web-owning female T. tenuis (0·71 ± 0·04 mg day−1) Ecological Society, Journal of Animal between the weight of field-collected female T. tenuis was found to be significantly greater than that for −1 Ecology, 72, and the number of prey items present within their web- non-web-owning individuals (0·29 ± 0·03 mg day )

745–756 site (Fig. 4), indicating that the weight of female T. tenuis (F1,302 = 38·53, P < 0·001). However, this translates 752 J. D. Harwood, K. D. Sunderland & W. O. C. Symondson

Fig. 3. Mean weight of Erigone atra, E. dentipalpis, Tenuiphantes tenuis and Bathyphantes gracilis captured from web-sites or non-web-sites. Bars are ± SE.

Fig. 4. Relationship between the number of prey at web-sites measured by web-centred quadrats and the weight of female Tenuiphantes tenuis captured at these sites. Data collected from four fields with varying prey availability (+, , , ). Regression: 2 y = 0·17x + 0·47 (r = 0·26, F1,136 = 25·86, P < 0·001).

into web-owning T. tenuis catching on average 5·2 D. This is the first study in which web-sites were paired melanogaster-sized prey items per week and non-web- directly with non-web-sites of comparable microhabi- owners just 2·1. Similarly, the mean estimated con- tat structure, providing a quantitative measure of web- sumption of male T. tenuis at web-sites (0·23 ± 0·02 mg site selection efficiency by spiders. We hypothesized day−1) was significantly greater than non-web-owning that, as their food supply is generally scarce (and also ± −11 males (0·13 0·01 mg day ) (F1,147 = 11·11, P < 0·01). variable in space and time) (Harwood et al. 2001), This equates to web-owning males consuming 1·7 D. linyphiids will have evolved behavioural mechanisms melanogaster-sized prey items per week and those for exploitation of prey-rich patches. Despite the fact without webs less than one. that there are many other factors affecting choice of web-locations (e.g. structural complexity of vegetation, microclimate and/or the avoidance of conspecifics and Discussion natural enemies: Samu, Sunderland & Szinetár 1999; In the Introduction we posed a series of hypotheses, Sunderland & Samu 2000; Symondson, Sunderland & interrelated to some degree, and we propose to discuss Greenstone 2002), our results show clearly that webs © 2003 British these directly. are significantly more likely to be found in areas of higher Ecological Society, Journal of Animal The hypothesis that spiders living within a relatively prey density. Even when comparisons were made between Ecology, 72, homogeneous ecosystem are able to locate their webs at web-sites and carefully matched non-web-sites, in terms 745–756 sites where prey are aggregated at the microhabitat level. of microhabitats, the number and biomass of prey 753 trapped at web-sites was 1·4 × (sticky traps) and 1·8 × density than at matched non-web-sites. However, Spider (quadrats) greater than at non-web-sites. Regardless Collembola were significantly more important to Erig- web-location of trapping system, Collembola were the main prey oninae, with nearly twice the biomass (sticky traps) of strategies that were apparently driving web-site selection (Fig. 1) Collembola trapped at their web-sites compared with and were consistently more numerous and of greater Linyphiinae. Linyphiinae rarely hunt for prey away biomass at web-sites. Field experiments manipulating from their webs (Alderweireldt 1994a), and fewer Col- Collembola density would be useful to test the strength of lembola would be expected within the webs of these these conclusions in an independent way. Quadrat data spiders, located above ground level, than in the webs also showed, however, that numbers and biomass of of Erigoninae. Nevertheless, Collembola density was most other prey were greater at web-sites, suggesting still higher at the web-sites of Linyphiinae than at non- that the spiders were not responding specifically to Col- web-sites. For Erigoninae, which regularly hunt away lembola but more generally to higher prey density. from their webs (Alderweireldt 1994a), Collembola Despite any dynamic intra- or interspecific interactions comprised a major component of available epigeal between spiders, therefore, which may have affected site prey. Overall, however, there was no significant differ- residence by individual spiders, the web-sites themselves ence in the abundance of prey at the web-sites of Eri- were associated positively with availability of prey. goninae compared with Linyphiinae, although the Over 95% of web-based linyphiids in our study con- biomass at the sites of Linyphiinae webs was lower. The structed their webs within rows of wheat (often around data show that there were in fact significantly greater the bases of wheat stems) rather than in the spaces densities of non-Collembolan prey at web-sites of Liny- between rows. Spiders are clearly not constructing their phiinae, compared with those of Erigoninae, mainly in webs solely on the basis of microhabitat structural the form of Aphididae and Thysanoptera. Nearly twice complexity, but are either selecting a prey-rich position as many aphids were recovered from the Linyphiinae or building at random and then showing greater web- web-sites. Aphids and Thysanoptera are more likely to site tenacity in prey-rich areas. The latter hypothesis is be intercepted by the aerial webs of the Linyphiinae likely given that spiders abandon patches or webs read- when these prey are either falling from above or flying. ily if their current location is not perceived as providing The hypothesis therefore appears to be supported: both sufficient numbers of prey (Janetos 1982; Olive 1982; subfamilies were locating their webs where prey density Persons & Uetz 1998). The elevated abundance and bio- was higher, but each was aggregating to microsites mass of prey at web-sites applied across a wide range where the prey spectrum present reflected the prey of prey taxa. Spiders in web-sites were therefore more most likely to be captured by their respective foraging likely to achieve a mixed diet, which translates in gen- strategies. eralist predators to improved survival, development The results from analyses addressing these first two and reproduction, and thus fitness (Greenstone 1979; hypotheses also argue against the hypothesis that the Toft & Wise 1999; Mayntz & Toft 2001). Why the prey spiders were simply responding to the same micro- are themselves aggregated, in microsites within the same habitat cues as their prey. If they were, then, assuming our microhabitats, may be due to many interacting factors matching of web-site microhabitats with selected non- (e.g. responses to aggregation pheromones, spatial web-sites was accurate, prey density should be the same clustering of offspring). Although we cannot exclude or lower (through patch depression) at web-sites. This entirely the possibility that some microhabitat variable proved not to be the case for either subfamily of liny- was causing such aggregation over a range of very dif- phiid; indeed, as reported above, prey densities were ferent taxa, even between careful matched microsites, it considerably greater at web-sites. Dynamic relocation within would appear more likely that the spiders were respond- their chosen microhabitats to patches with greater ing directly to prey aggregations. In a later study, for abundance of the types of prey most suited to their for- example, we could find no significant difference in aging strategies would be the most likely explanation. the temperature between web and non-web-sites visu- The high levels of competition for web-sites, reported in ally matched for microsite similarity (Sunderland, the literature (Samu et al. 1996; Heiling & Herberstein unpublished data). 1999; Riechert & Hall 2000), might suggest that a dom- The hypothesis that aggregation to prey density would inant spider looking for a good web-site would find one be determined by the types of prey captured most readily more rapidly by serially displacing spiders from exist- by the contrasting foraging strategies of two spider sub- ing web-sites rather than by random selection of a site families. Earlier we proposed that there were clear that had the right structure and microclimate. The parallels between spider web-sites and plants in hay costs of competition can be high (the loser might be meadows. The foraging strategies of different species of eaten) but in an environment where most spiders lack- plant allow them to exploit different resources (e.g. ing a good web-site were starving the risks may be deep/shallow rooted species), avoiding direct competi- worth taking to ensure higher fecundity. © 2003 British tion and hence allowing many species to coexist. So The hypothesis that the different microsites, to which Ecological Society, Journal of Animal too, it seems, with spiders. Separate analyses of prey at the different spider subfamilies were responding, would Ecology, 72, web-sites of Linyphiinae and Erigoninae showed that show differences in the relative abundance of different 745–756 both subfamilies selected microsites with greater prey types of prey. We proposed (above) that spiders may 754 simply build their webs in response to microhabitat- ing period (0·12 mg prey cm−2 day−1) corresponded to J. D. Harwood, defined criteria, then move to new sites if prey avail- approximately one D. melanogaster-sized prey item K. D. Sunderland & ability proved to be inadequate. If this is correct then it entering a web every day (cf. feeding regime 1, Fig. 2), W. O. C. Symondson is microhabitat selection that defines the spectrum of which is comparable to feeding rates for agricultural prey present. Comparison of prey density and type at spiders in Europe and the United States (Nyffeler & non-web-sites, that were matched to the web-sites of Sunderland 2003). Consumption by non-web-owning the two subfamilies but not chosen by them as web- T. tenuis was less than half that of T. tenuis collected sites, should reflect a similar pattern of the ratios of dif- from webs, indicating severe starvation. ferent types of prey to those at web-sites (discussed The measurement of T. tenuis weight from labor- above). This proved to be the case, suggesting strongly atory feeding trials indicated a linear relationship with

that the two subfamilies were building their webs in log10 prey biomass consumed. This approach to estimation microsites best suited to finding the prey categories of prey consumption in the field may provide a more caught most readily by their contrasting hunting strat- rapid model for determining the condition of field- egies. There were more Collembola at the non-web- collected predators than previously applied techniques sites of Erigoninae and more Aphididae and Thysan- (Anderson 1974; Wise 1979; Juliano 1986; Jakob, Marshall optera at the non-web-sites of Linyphiinae (numbers & Uetz 1996; Bilde & Toft 1998). It is, however, import- and biomass). These results confirm that our matching ant to remember that the model would change if tem- of non-web-sites to web-sites must have reflected accu- peratures were higher or lower and/or food quality was rately microsite selection by the two spider subfamilies, better or worse than Drosophila. and that the microhabitats chosen were significantly It was notable in our study that pest biomass formed different in ways that were reflected in the prey spectra a small proportion of the total prey available to the that were found within them. spiders. Non-pest prey (such as Collembola, Diptera The hypothesis that locating webs where prey density and Hymenoptera) are often of greater nutritional value was higher results in an improved nutritional status for than pests for spiders (Toft 1999), and can make a major the spiders and that this would affect the web-dependent contribution to supporting predator population growth Linyphiinae more strongly than the less web-dependent (Sigsgaard, Toft & Villareal 2001) which, in turn, Erigoninae. We compared the nutritional status of spi- increases the contribution that spiders make to pest ders at web-sites with those captured away from webs. control (Settle et al. 1996). T. tenuis was the most abundant Spider weight was used as a surrogate for nutritional linyphiid captured, and could be the most important status (Anderson 1974; Nakamura 1977), which is reason- in terms of biological control due to the construction able given the strong relationship between the rate of large webs which intercept considerable numbers of at which prey were consumed and spider weight (Fig. 2). falling aphids. In addition, the aphids’ low escape fre- Web-owning female and male T. tenuis were consider- quency (Carter et al. 1982) suggests that these spiders ably heavier (45% and 18%, respectively) than spiders could contribute significantly to limiting pest popula- without webs (Fig. 3), a result supported by a previous tions. This would be especially important when con- year’s data (41% and 21%, respectively). This suggests sidering the potential for exploiting the switching that there is a higher degree of web-dependence by this capacity of generalist predators (Symondson et al. species than by any of the Erigoninae. Male spiders 2002) by modifying farming practices to boost the of most species are found away from their webs more detrital food chain early in the season (Wise, Snyder & often than females, but this relates to higher male activ- Tuntibunpakal 1999; Sunderland & Samu 2000). ity, especially mate finding (Alderweireldt 1994b; In conclusion, differences in foraging strategies by Anderson & Morse 2001). There was a highly signi- two subfamilies of Linyphiidae resulted in aggregation ficant relationship between female T. tenuis weight and to microsites with significantly different spectra of prey. availability of prey within microhabitats across four Uetz, Halaj & Cady (1999), while allocating spiders to fields with very different overall levels of prey abund- guilds on the basis of their foraging strategies, made no ance (Fig. 4). Thus, for the web-dependent T. tenuis at allowance for differences below the family level. Our least, web-location at sites with higher prey availability data suggest that Linyphiinae and Erigoninae may not resulted in improved nutritional status. When T. tenuis substitute for one another within the same taxonomic weights were converted to numbers of prey consumed or functional guild, but rather, by avoiding direct com- it was revealed that females in webs may be consuming petition through a degree of resource partitioning, can ∼ 2·5 × as much prey biomass as non-web-owners, and coexist. It may well be that as we increase our know- even male web-owners were consuming nearly twice as ledge of generalist predator foraging behaviour and much as those found away from their web-sites. It is not prey choice, through detailed field studies such as this, too surprising that the less web-dependent Erigoninae or by using molecular techniques to analyse food webs did not show such a relationship; many of the Erigo- (Symondson 2002), the elegance with which appar- © 2003 British ninae classed as non-web-owners may indeed have been ently functionally similar species intermesh within bio- Ecological Society, Journal of Animal web-owners, but were simply captured when hunting logically diverse ecosystems, within their own unique Ecology, 72, away from their web-sites. The mean biomass of prey and often narrow niche axes may undermine current 745–756 available to Erigoninae during the 3-month monitor- attempts to ‘lump’ species within guilds. 755 Acknowledgements Nakamura, K. (1977) A model for the functional response of Spider a predator to varying prey densities; based on the feeding We are very grateful to John Fenlon (Horticulture web-location ecology of wolf spiders. Bulletin of the National Institute of Research International), who provided valuable statis- Agricultural Sciences, 31, 29–89. strategies tical advice. Keith Sunderland was funded by the Nyffeler, M. & Sunderland, K.D. (2003) Composition, abund- ance and pest control potential of spider communities Department for Environment, Food and Rural Affairs. in agroecosystems: a comparison of European and US This research was supported by grants from Horti- studies. Agriculture Ecosystems and Environment, 95, 579– culture Research International and Cardiff University. 612. Olive, C.W. (1982) Behavioural response of a sit-and-wait predator to spatial variation in foraging gain. Ecology, 63, References 912–920. Persons, M.H. & Uetz, G.W. (1998) Presampling sensory Adams, M.R. (2000) Choosing hunting sites: web-site prefer- information and prey density consumption by wolf spiders ences of the orb weaving spider, Neoscona crucifera, relative (Araneae, Lycosidae). Behavioural Ecology, 9, 360–366. to light cues. Journal of Insect Behaviour, 13, 299–305. Pierce, G.J. & Ollason, J.G. (1987) Eight reasons why optimal Alderweireldt, M. (1994a) Prey selection and prey capture foraging theory is a complete waste of time. Oikos, 49, 111– strategies of linyphiid spiders in high-input agricultural fields. 117. Bulletin of the British Arachnological Society, 9, 300–308. Pyke, G.H. (1984) Optimal foraging theory: a critical Alderweireldt, M. (1994b) Day/night activity rhythms of review. Annual Review of Ecology and Systematics, 15, spiders occurring in crop-rotated fields. European Journal 523–575. of Soil Biology, 30, 55–61. Riechert, S.E. (1992) Spiders as representative ‘sit-and-wait’ Anderson, J.F. (1974) Responses to starvation in the spiders predators. Natural Enemies: the Population Biology of Lycosa lenta Hentz and Filistata hibernalis (Hentz). Ecology, Predators, Parasites and Diseases (ed. M.J. Crawley), 55, 576–585. pp. 313–328. Blackwell Scientific Publications, Oxford, Anderson, J.T. & Morse, D.H. (2001) Pick-up lines: cues UK. used by male crab spiders to find reproductive females. Riechert, S.E. & Hall, R.F. (2000) Local population success Behavioural Ecology, 12, 360–366. in heterogeneous habitats: reciprocal transplant experi- Bilde, T. & Toft, S. (1998) Quantifying food limitation of ments completed on a desert spider. Journal of Evolutionary arthropod predators in the field. Oecologia, 115, 54–58. Biology, 13, 541–550. Brown, K.M. (1981) Foraging ecology and niche partitioning Samu, F., Sunderland, K.D. & Szinetár, C. (1999) Scale- in orb-weaving spiders. Oecologia, 50, 380–385. dependent dispersal and distribution patterns of spiders in Carter, N., Gardner, S., Fraser, A.M. & Adams, T.H.L. agricultural systems: a review. Journal of Arachnology, 27, (1982) The role of natural enemies in cereal aphid popu- 325–332. lations. Proceedings of the Entomology Group of the Samu, F., Sunderland, K.D., Topping, C.J. & Fenlon, J.S. Association of Applied Biologists − Natural Enemies and Insect (1996) A spider population in flux: selection and abandon- Population Dynamics. Annals of Applied Biology, 101, ment of artificial web-sites and the importance of inter- 190–195. specific interactions in Lepthyphantes tenuis (Araneae: Charnov, E.L. (1976) Optimal foraging: the marginal value Linyphiidae) in wheat. Oecologia, 106, 228–239. theorem. Theoretical Population Biology, 9, 129–136. Settle, W.H., Ariawan, H., Astuti, E.T., Cahyana, W., Hakim, A.L., Crawley, M.J. (1997) The structure of plant communities. Hindayana, D., Lestari, A.S. & Sartanto, P. (1996) Manag- Plant Ecology, 2nd edn (ed. M.J. Crawley), pp. 475–531. ing tropical rice pests through conservation of generalist Blackwell Science, Oxford, UK. natural enemies and alternative prey. Ecology, 77, 1975– Gillespie, R.G. & Caraco, T. (1987) Risk-sensitive foraging 1988. strategies of two spider populations. Ecology, 68, 887–899. Sigsgaard, L., Toft, S. & Villareal, S. (2001) Diet-dependent Greenstone, M.H. (1979) Spider feeding behaviour optimises survival, development and fecundity of the spider Atypena dietary essential amino acid composition. Nature, 282, formosana (Oi) (Araneae: Linyphiidae) − implications for 501–503. biological control in rice. Biocontrol Science and Technology, Harwood, J.D., Sunderland, K.D. & Symondson, W.O.C. 11, 233–244. (2001) Living where the food is: web location by linyphiid Silverton, J. & Charlesworth, D. (2001) Introduction to Plant spiders in relation to prey availability in winter wheat. Population Biology, 4th edn. Blackwell Science, Oxford, Journal of Applied Ecology, 38, 88–99. UK. Heiling, A.M. & Herberstein, M.E. (1999) The importance of Smallwood, P.D. (1993) Web-site tenure in the long-jawed being larger: intraspecific competition for prime web-sites spider − it is risk-sensitive foraging or conspecific interactions. in orb-web spiders (Araneae, Araneidae). Behaviour, 136, Ecology, 74, 1826–1835. 669–677. Stephens, D.W. & Krebs, J.R. (1986) Foraging Theory. Jakob, E.M., Marshall, S.D. & Uetz, G.W. (1996) Estimating Princeton University Press, Princeton, USA. fitness: a comparison of body condition indices. Oikos, 77, Sunderland, K.D., Chambers, R.J. & Carter, O.C.R. (1988) 61–67. Potential interactions between varietal resistance and nat- Janetos, A.C. (1982) Foraging tactics of two guilds of web- ural enemies in the control of cereal aphids. Integrated Crop spinning spiders. Behavioural Ecology and Sociobiology, 10, Protection in Cereals (eds R. Cavalloro & K.D. Sunderland), 19–27. pp. 41–56. A.A. Balkema, Rotterdam, The Netherlands. Juliano, S.A. (1986) Food limitation of reproduction and sur- Sunderland, K.D., Fraser, A.M. & Dixon, A.F.G. (1986) vival for populations of Brachinus (Coleoptera: Carabidae). Distribution of linyphiid spiders in relation to capture of prey Ecology, 67, 1036–1045. in cereal fields. Pedobiologia, 29, 367–375. Leclerc, J. (1991) Optimal foraging strategy of the sheet-web Sunderland, K.D. & Samu, F. (2000) Effects of agricultural © 2003 British spider Lepthyphantes flavipes under perturbation. Ecology, diversification on the abundance, distribution, and pest control Ecological Society, 72, 1267–1272. potential of spiders: a review. Entomologia Experimentalis Journal of Animal Mayntz, D. & Toft, S. (2001) Nutrient composition of the et Applicata, 95, 1–13. Ecology, 72, prey’s diet affects growth and survivorship of a generalist Symondson, W.O.C. (2002) Molecular identification of prey 745–756 predator. Oecologia, 127, 207–213. in predator diets. Molecular Ecology, 11, 627–641. 756 Symondson, W.O.C., Sunderland, K.D. & Greenstone, M.H. in cereals, with a simple method for estimating percentage J. D. Harwood, (2002) Can generalist predators be effective biocontrol parasitism of Erigone spp. eggsacs by Hymenoptera. K. D. Sunderland & agents? Annual Review of Entomology, 47, 561–594. Journal of Applied Entomology, 118, 217–223. Toft, S. (1999) Prey choice and spider fitness. Journal of Wise, D.H. (1979) Effects of an experimental increase in prey W. O. C. Symondson Arachnology, 27, 301–307. abundance upon the reproductive rates of two orb-weaving Toft, S. & Wise, D.H. (1999) Growth, development and sur- spider species (Araneae: Araneidae). Oecologia, 41, 289– vival of a generalist predator fed single- and mixed-species 300. diets of different quality. Oecologia, 119, 191–197. Wise, D.H. (1993) Spiders in Ecological Webs. Cambridge Uetz, G.W. (1992) Foraging strategies of spiders. Trends in University Press, Cambridge, UK. Ecology and Evolution, 7, 155–159. Wise, D.H., Snyder, W.E. & Tuntibunpakul, P. (1999) Spiders Uetz, G.W., Halaj, J. & Cady, A.B. (1999) Guild structure of in decomposition food webs of agroecosystems: theory and spiders in major crops. Journal of Arachnology, 27, 270–280. evidence. Journal of Arachnology, 27, 363–370. Van Baarlen, P., Sunderland, K.D. & Topping, C.J. (1994) Eggsac parasitism of money spiders (Araneae, Linyphiidae) Received 3 February 2003; accepted 11 April 2003

© 2003 British Ecological Society, Journal of Animal Ecology, 72, 745–756